Evaluating ChatGPT's Ability to Answer Common Parent Questions on Avascular Necrosis in Children

Area of Science:

• Orthopedics
• Medical Education
• Artificial Intelligence in Healthcare

Background:

• Avascular necrosis (AVN) causes significant pain, prompting patients and families to seek reliable information.
• The increasing prevalence of AI chatbots like ChatGPT necessitates an evaluation of their utility for medical information dissemination.

Purpose of the Study:

• To assess expert clinicians' perceptions of ChatGPT's responses to common pediatric avascular necrosis (AVN) questions.
• To evaluate the readability and suitability of ChatGPT-generated AVN information for parents.
• To gauge provider perspectives on AI's role in patient education for AVN.

Main Methods:

• A cross-sectional survey of 9 pediatric orthopedic and oncology specialists.
• ChatGPT was prompted to answer 15 common parent questions about AVN at a sixth-grade reading level.
• Responses were evaluated by clinicians for quality and clarity, and assessed for readability using Flesch-Kincaid and Gunning Fog indices.

Main Results:

• Clinicians rated all 15 ChatGPT responses as satisfactory, needing minimal clarification.
• Only 20% of responses met the requested sixth-grade reading level; average readability was between 7th grade and early high school.
• 100% of providers found the information at an appropriate reading level, and 56% believed it sufficient for parents.

Conclusions:

• ChatGPT can be a valuable resource for patient and family education regarding avascular necrosis (AVN).
• While responses were generally satisfactory, readability may not consistently meet specified levels, indicating a need for careful review.
• Further research is warranted to explore AI's role and limitations in providing medical information to patients.